Might there be a non-linear relationship between inequality and growth? A new paper suggests so. It finds that "there is a range of values of changes in inequality where the link [with GDP growth] is weak", but that big reductions in inequality do boost growth. It concludes:

Not all polices that reduce inequality will lead to faster economic growth. Instead, only those that greatly reduce inequality will.

The paper is purely statistical, which poses the question: what sort of mechanisms might generate such a pattern?

One possibility is that inequality depresses growth by generating a culture hostile to prosperity. For example, it might lead to mutual distrust (pdf) which is bad for growth, or to learned helplessness among the poor which saps their energy - say, to stay on in education or to take more initiative at work. It requires a big change in inequality to remove these cultural obstacles to growth. A change in the Gini coefficient of one or two percentage points doesn't much change culture.

A second possibility is that inequality is bad for growth because it imposes credit constraints upon workers which prevents them from forming worker coops even though it might be efficient to do so. A small reduction in inequality might not be sufficient to overcome these constraints.

Thirdly, whilst modest redistribution might have some benefits - for example in shifting incomes from those with a low propensity to consume to those with a higher - it can also have an adverse effect. If it leads to expectations of further redistribution, it could depress investment as capitalists anticipate low post-tax returns. However, a big redistribution might not have such ill-effects if it reduces the risk of future further tax rises by, in effect, buying off discontent. This is what James Buchanan had in mind when he wrote:

The rich man, who may sense the vulnerability of his nominal claims in the existing state of affairs and who may, at the same time, desire that the range of collective or state action be restricted, can potentially agree on a once-and-for-all or quasi-permanent transfer of wealth to the poor man, a transfer made in exchange for the latter's agreement to a genuinely new constitution that will overtly limit governmentally directed fiscal transfers. (The Limits of Liberty, 7.10.40)

He wrote that in the mid-70s - a time when the threat of redistribution might well have been depressing investment and growth.

Now, I'm not saying all this to say that radical redistribution is definitely a good thing; that requires a lot more arguing. I do so instead to challenge a common prior among social democrats - which might well arise from a presumption that the economy is a simple linear system - that mild reformism and "moderation" is sensible and "realistic" whereas radicalism is economically unsound. It ain't necessarily so.